From Halpern et al. 2012 supplemental info:
The ‘Lasting Special Places’ sub-goal focuses instead on those geographic locations that hold particular value for aesthetic, spiritual, cultural, recreational or existence reasons57. This sub-goal is particularly hard to quantify. Ideally one would survey every community around the world to determine the top list of special places, and then assess how those locations are faring relative to a desired state (e.g., protected or well managed). The reality is that such lists do not exist. Instead, we assume areas that are protected represent these special places (i.e. the effort to protect them suggests they are important places).
Clearly this is an imperfect assumption but in many cases it will be true. Using lists of protected areas as the catalogue of special places then creates the problem of determining a reference condition. We do not know how many special places have yet to be protected, and so we end up having all identified special places also being protected. To solve this problem we make two important assumptions. First, we assume that all countries have roughly the same percentage of their coastal waters and coastline that qualify as lasting special places. In other words, they all have the same reference target (as a percentage of the total area). Second, we assume that the target reference level is 30% of area protected.
The model for this goal considers the inland coastal zone (up to 1 km inland) independently from, and equally weighted with, the offshore coastal zone (up to 3 nm offshore). The status for this goal is calculated as:
\[X_{LSP} = \frac{\left(\frac{Area_{P}}{Area_{P_{ref}}} + \frac{Area_{MPA}}{Area_{MPA_{ref}}}\right)}{2}\]
where:
New World Database on Protected Areas spatial data from UNEP-WCMC became available in May 2016. In addition, an error in the previous assessments (specifically the .csv used to determine total areas for the 3 nm offshore and 1 km inland regions) caused total region areas to be underestimated, artificially inflating scores in past assessments, in some cases quite significantly. This error has been fixed in the v2016 assessment, but in so doing, many of the reported scores are noticeably lower than before.
Reference: IUCN and UNEP-WCMC (2016), The World Database on Protected Areas (WDPA) [On-line], May 2016. Cambridge, UK: UNEP-WCMC. Available at: www.protectedplanet.net.
Downloaded: June 7, 2016
Description: Shapefile of World Database on Protected Areas
Time range: 1819 - 2015; some protected areas do not have an associated “status year” and are reported as year 0.
Format: Shapefile
File location: Mazu:git-annex/globalprep/_raw_data/wdpa_mpa/d2016/WDPA_May2016-shapefile/
The WDPA-MPA dataset comes as a shapefile or geodatabase in WGS84 coordinate reference system.
Once reprojected to Mollweide, the WDPA polygons are rasterized to a 500 m grid. This resolution is necessary to reasonably capture the “signal” of a 1000 m feature, i.e. the 1 km inland coastal zone. Due to the fine scale, we continue to rely on ArcGIS functionality to perform this rasterization. For future years we hope to perform this in an open-source system such as GDAL.
rasterize_wdpa.py script is an ArcPy script, and therefore, unfortunately, needs to be run on a Windows system with access to ArcGIS.rasterize_wdpa.py script relies on an existing 500 m Mollweide raster to set extents and resolution. Either of these rasters (for the 3 nautical mile offshore and 1 km inland regions) can be used:
git-annex/globalprep/spatial/v2015/data/rgn_raster_500m/rgn_inland1km_mol_500mcell.tifgit-annex/globalprep/spatial/v2015/data/rgn_raster_500m/rgn_offshore3nm_mol_500mcell.tifrasterize_wdpa.py is saved as a 500 m global Mollweide raster, in which each cell value indicates the earliest year of protection.
git-annex/globalprep/lsp/v2016/int/wdpa_designated_mol.tifComparing the global WDPA raster to the 3 nm offshore and 1 km inland rasters, we can tally the protected area within each region and compare to the total area within each region. Note each cell is 500 m2, so area is .25 km2, but since we are simply calculating a ratio, this cancels out.
Once the WDPA raster is cross-tabulated against the OHI region rasters (both 3 nm offshore and 1 km inland) we have the number of protected cells, identified by year of protection, within each region. NA values are unprotected cells.
## year rgn_id n_cells
## Min. : 0 Min. : 1.0 Min. : 0
## 1st Qu.:1905 1st Qu.: 59.0 1st Qu.: 0
## Median :1942 Median :117.5 Median : 0
## Mean :1928 Mean :118.8 Mean : 604
## 3rd Qu.:1978 3rd Qu.:178.0 3rd Qu.: 0
## Max. :2015 Max. :272.0 Max. :2365009
## NA's :222
## year rgn_id n_cells
## Min. : 0 Min. : 1.0 Min. : 0.0
## 1st Qu.:1905 1st Qu.: 59.0 1st Qu.: 0.0
## Median :1942 Median :117.0 Median : 0.0
## Mean :1928 Mean :118.1 Mean : 137.7
## 3rd Qu.:1978 3rd Qu.:177.0 3rd Qu.: 0.0
## Max. :2015 Max. :250.0 Max. :639110.0
## NA's :221
## year rgn_id n_cells
## Min. : 0 Min. : 1.00 Min. : 0
## 1st Qu.:1905 1st Qu.: 62.75 1st Qu.: 0
## Median :1942 Median :124.50 Median : 0
## Mean :1928 Mean :127.64 Mean : 70979
## 3rd Qu.:1978 3rd Qu.:188.25 3rd Qu.: 0
## Max. :2015 Max. :277.00 Max. :994469137
## NA's :237 NA's :149
Grouping by rgn_id, the total number of cells per region is determined by summing cell counts across ALL years, including cells with year == NA (unprotected cells). We can then determine the protected area for each year by looking at the cumulative sum of cells up to any given year.
Since the cells are 500 m on a side, we can easily calculate area by multiplying cell count * 0.25 km2 per cell.
Finally we can calculate the status of a region for any given year by finding the ratio of protected:total and normalizing by the goal’s target of 30% protected area.
The status is based on a simple arithmetic average of the inland and offshore status values.
We can save outputs for the following layers:
~/github/ohiprep/globalprep/lsp/v2016/output/lsp_protected_inland1km.csv: inland protected area (km2) for each region (since 2000)~/github/ohiprep/globalprep/lsp/v2016/output/lsp_protected_offshore3nm.csv: offshore protected area (km2) for each region (since 2000)~/github/ohiprep/globalprep/lsp/v2016/output/lsp_a_total_inland1km.csv: inland 1 km total area (km2) for each region~/github/ohiprep/globalprep/lsp/v2016/output/lsp_a_total_offshore3nm.csv: offshore 3 nm total area (km2) for each region~/github/ohiprep/globalprep/lsp/v2016/output/lsp_protected_eez.csv: EEZ protected area (km2) for each region (since 2000) (for resilience)~/github/ohiprep/globalprep/lsp/v2016/output/lsp_a_total_eez.csv: EEZ total area (km2) for each region (for resilience)From the inland and offshore layers, we can also estimate the status and trend. “Official” values will be determined in the toolbox.
Year-by-year status and trend estimates will be saved:
~/github/ohiprep/globalprep/lsp/v2016/output/lsp_status.csv: estimate of status by region since 2000~/github/ohiprep/globalprep/lsp/v2016/output/lsp_trend.csv: estimate of trend by region since 2010One key difference in the v2016 process is the update to the region total areas. Prior assessments used a .csv of areas for 1km inland and 3 nm offshore buffers; unfortunately this file seems to have had major flaws. The new method calculates the buffer areas directly from the rasters. The results compare very closely to areas calculated with rgeos::gArea() and to the putative region areas within the shapefile itself, but very different from those values in the .csv.
This large shift in actual total area (generally much larger than previously reported) has the effect of significantly reducing the ratio of protected:total area, thus lowering the region LSP status scores in general below previously reported scores.